The dose response of noncancer disease mortality differs by time period for atomic-bomb survivors of the Life Span Study (LSS) in the Radiation Effects Research Foundation. We applied change point models and Akaike’s Information Criteria to the data of LSS Report 13 (Preston et al. (2003)), including 86,572 subjects and 31,881 noncancer disease deaths during 1950 to 1997. A model which incorporates a change point in the dose response but not in the background model showed results similar to those in LSS Report 13, which suggested that the baseline rate difference between proximal and distal survivors changed by time and the preferred dose response might be linear-quadratic in 1950-1967 and linear in 1968-1997. However, the present model, which incorporates change points in both dose response and background models, showed that there was, little evidence to support a time-dependent change in the baseline rate difference between proximal and distal survivors. The preferred dose response was pure-quadratic in 1950-1964 and linear in 1965-1997. When the data were divided into cardiovascular disease (CVD) mortality and noncancer disease mortality other than CVD, the shape of the dose response did not differ by time period (linear for CVD and pure-quadratic for other than CVD).
The importance of cohort studies in order to explore diseae causes has been recognized. When a disease is rare, the cohort studies need a long follow-up period, a high budget, and a large number of participants in the target population. A nested case-control study design is proposed as a supplementary study conducted within a cohort, efficiently using the information from the cohort study. In the present paper we explain the sampling design of controls and the methods of data analysis in a nested case-control study. In addition we explain the methods of sample size and power calculation to design the nested case-control study. Further, we introduce some methods for the data with missing values in covariates, and we report the performance assessment from the application of these methods to the nested case-control study data.
This paper reviews basic ideas of Structural Causal Models (SCMs) proposed by Judea Pearl (1995, 2009a). SCMs are nonparametric structual equation models which express cause-effect relationship between variables, and justify matematical principles of both the potential outcome approach and the graphical model approach for statistical causal inference. In this paper, considering the difference/connection between SCMs and Rubin's Causal Models (RCMs) (Rubin, 1974, 1978, 2006), we state that (1) the expressive power of the potential outcome approach is higher than that of the graphical model approach, but (2) the graphical model approach. From these consderations, we conclude that we should discuss statistical causal inference based on both approaches.